def results(self):
while 0 < get_score(self.dealer_cards) < 17 or is_soft_17(
self.dealer_cards):
self.dealer_cards.append(self.deck.next_card())
self.dealer_score = get_score(self.dealer_cards)
for i in range(self.number_of_players):
current = []
for j in range(len(self.hands[i])):
current.append(self.evaluate(i, j))
self.hand_rewards[i] = current
shuffle = False
if self.deck.index > self.shuffle_every:
shuffle = True
return {
"shuffle": shuffle,
"done": True,
"dealer_cards": self.dealer_cards,
"dealer_score": self.dealer_score,
"hands": self.hands,
"rewards": self.hand_rewards,
"count": self.deck.counter.get_rc()
}
def simple_play_2(dealer, strategy):
res = dealer.reset()
while not res["done"]:
player_playing = res["player_playing"]
hand_playing = res["hand_playing"]
dealer_cards = res["dealer_cards"]
hand = res["hands"][player_playing][hand_playing]
can_split = len(res["hands"][player_playing]) == 1
action = choose_action(hand,
dealer_cards,
strategy=strategy,
can_split=can_split)
res = dealer.step(action)
player_blackjacks = [get_score(cards) for cards in res["hands"][0]]
player_blackjacks = [s == 22 for s in player_blackjacks]
player_blackjacks = [1 if s else 0 for s in player_blackjacks]
dealer_blackjack = 1 if get_score(res["dealer_cards"]) == 22 else 0
dealer_burst = 1 if get_score(res["dealer_cards"]) == 0 else 0
rewards = res["rewards"][0]
return np.mean(rewards), np.mean(
player_blackjacks), dealer_blackjack, dealer_burst
def test_get_score(self):
cards = [[1, 5, 5], [1, 3, 8], [2, 3, 8], [10, 1], [1, 10], [1, 10, 4],
[4, 5, 12, 10], [13, 4], [1, 1, 1, 1, 1]]
truth = [21, 12, 13, 22, 22, 15, 0, 14, 15]
scores = [get_score(c) for c in cards]
print(scores)
for i in range(len(truth)):
self.assertEqual(scores[i], truth[i])
def evaluate(self, player_id, hand_id):
doubled = hand_id in self.doubled_hands[player_id]
x = 2 if doubled else 1
player_score = get_score(self.hands[player_id][hand_id])
if player_score == 22 and len(self.hands[player_id]) >= 2:
player_score = 21
if player_score == 0:
return -x
if player_score == 22 and self.dealer_score < 22:
return 1.5 * x
if self.dealer_score > player_score:
return -x
if self.dealer_score == player_score:
return 0
return x
def reset(self):
self.shuffle_if_needed()
self.hands = {}
self.dealer_cards = []
self.doubled_hands = {}
self.hand_rewards = {}
self.dealer_score = None
for i in range(self.number_of_players):
self.doubled_hands[i] = []
self.hand_rewards[i] = []
self.player_playing = 0
self.hand_playing = 0
for i in range(self.number_of_players):
self.hands[i] = [[self.deck.next_card()]]
self.dealer_cards.append(self.deck.next_card())
for i in range(self.number_of_players):
# TODO: check that this is the way to do this
self.hands[i][0].append(self.deck.next_card())
# pour le blackjack américain il faut rajouter ça ici:
# self.dealer_cards.append(self.deck.next_card())
for i in range(self.number_of_players):
self.player_playing = i
if get_score(self.hands[i][0]) not in [0, 22]:
done = False
break
else:
done = True
if self.player_playing == self.number_of_players - 1 and done:
return self.results()
return {
"done": False,
"dealer_cards": self.dealer_cards,
"hands": self.hands,
"hand_playing": self.hand_playing,
"player_playing": self.player_playing
}
def play(n):
"""
play n games of blackjack
"""
for i in range(n):
print("\nNew game:")
res = dealer.reset()
while not res["done"]:
print("hands:")
print(res["hands"])
print("dealer:")
print(res["dealer_cards"])
player_playing = res["player_playing"]
hand_playing = res["hand_playing"]
hand = res["hands"][player_playing][hand_playing]
score = get_score(hand)
print("hand playing: "+str(hand))
print("your score: "+str(score))
action = ""
while action not in action_space:
action = input("please enter an action\n")
res = dealer.step(action)
print(res)
def hit(self, player_id, hand_id):
self.hands[player_id][hand_id].append(self.deck.next_card())
score = get_score(self.hands[player_id][hand_id])
# TODO: remplacer par score in [0, 21, 22] ?
return score in [0, 22]
def naive_strategy(hand, dealer_hand, strategy):
if get_score(hand) > strategy["k"]:
return "stick"
return "hit"